Purification and Properties of Thiol ,8-Lactamase A MUTANT OF pBR322 p-LACTAMASE IN WHICH THE ACTIVE SITE SERINE HAS BEEN REPLACED WITH CYSTEINE*

The specifically mutated enzyme thiol p-lactamase has been expressed in Escherichia coli by means of the trp promoter and purified to homogeneity. The gene for this enzyme results from a single base change N4lo A + T in the gene of pBR322 RTEM p-lactamase (EC 3.5.2.6, penicillinase, penicillin amido-p-lactamhydro-lase) which alters the codon for the active site Ser 70 to that for Cys. Precursor thiol /3-lactamase is proc-essed to give the same NHz-terminal sequence as that for wild type enzyme. In contrast to the wild type enzyme, thiol 8-lactamase contains one free titratable thiol group/molecule. Thiol 6-lactamase catalyzes the hydrolysis of p-lactams with a substrate specificity that is distinct from that of wild type enzyme. For benzyl- penicillin and ampicillin, the K , values are similar to wild type values although the kcat values are 1-2% that of wild type enzyme. For the cephalosporin nitrocefin, the K , is greater than 10-fold that of the wild type and the kcat is at least as large as the kcat for the wild type enzyme. Thiol p-lactamase is different from wild type p-lactamase in that it is not competitively inhibited by

that occur at the active site of hydrolytic enzymes (7). a-Lactamases hydrolyze the 8-lactam amide bond of antibiotics such as penicillins and cephalosporins and are the most common reason for bacterial resistance to these drugs (8). Understanding the mechanism of these enzymes is pertinent to the design of pharmaceutically important inhibitors (9). Of the three classes of p-lactamases (A, B, and C), A and C are serine hydrolytic enzymes (10). Recent studies, mainly utilizing poor substrates, have shown that for these enzymes, a serine residue (Ser 70 for Class A and Ser 80 for Class C) acts as an active site nucleophile with the formation of an acyl-enzyme intermediate (11)(12)(13)(14)(15). As a result of this intermediate, product partitioning occurs for the Class C enzymes when alcohols are present (15).
For one Class A enzyme, RTEM p-lactamase, the actual acyl-enzyme intermediate was spectrophotometrically observed during the slow hydrolysis of the inhibitor cefoxitin (11). These serine p-lactamases together with their homologous D-alanylcarboxypeptidases do not appear to possess a conserved histidine residue and thus they form a family of serine hydrolytic enzymes distinct from proteases like subtilisin (16).
While both serine and cysteine residues can be effective nucleophiles (both serine and cysteine proteases occur in nature), the active site hydroxyl and sulfhydryl groups do have unique chemical and physical properties. The replacement of an active site serine by a cysteine residue, a net substitution of an-OH group by an -SH group, probes the mechanism of catalysis by perturbing in a precise and significant manner the chemical environment of the active site.
Chemical methods were previously used to replace the -OH group of subtilisin's active site serine by -SH (17,18). We substituted the Ser 70' of pBR322 RTEM p-lactamase (EC 3.5.2.6, (penicillinase, penicillin amido-P-lactamhydrolase) with Cys by recombinant DNA methods. Specifically, by making a single base mutation in the gene for pBR322 8lactamase, we changed the codon for Ser 70 AGC to TGC (Cys). The new enzyme, thiol 8-lactamase, was active in uiuo, and its i n vitro 0-lactamase activity was sensitive to mercuric ion. We report here the purification of this enzyme and describe some of its enzymatic and physical chemical properties. Thiol 8-Lactamase: Purification and Properties the single base substitution N"' T + A' and codes for thiol plactamase (7). Plasmid pOTBL is a derivative of pIS2 which contains the trp promoter at the ClaI site orientated in the direction of the thiol 8-lactamase gene (21)'.
General Methods-Proteins were separated by polyacrylamide slab gel electrophoresis in the presence of sodium dodecyl sulfate as described by Laemmli (23). Protein samples were denatured by boiling 5-10 min in sample buffer (50 mM Tris/HCl, pH 7.0, 4% sodium dodecyl sulfate, 1.0 M 2-mercaptoethanol, 15% (w/v) glycerol, 0.001% bromphenol blue). Protein bands were detected by staining with Coomassie brilliant blue. General protein concentrations were measured by the Bio-Rad assay. Standard microbiological (24) and recombinant DNA methods (25) were utilized. Cell-free extracts were prepared as previously described (7). High pressure liquid chromatography was performed using a DuPont Instruments Series 880 high pressure liquid chromatographer equipped with a DuPont Instruments variable wavelength detector.
Expression-Induction of the trp promoter in strain pOTBL/ HBlOl occurred in minimal media (24) (M9 salts, 2% glucose, 0.2% casamino acids) at 37 "C. Cells were grown to late log phase in LB media (24) containing 5 mg/liter of ampicillin, centrifuged, washed in minimal media, and resuspended to a final A595 of 0.5. Expression of thiol 8-lactamase was monitored by the enzymatic activity in both the media and in cell-free extracts of aliquots and by SDS3 electrophoresis of whole cell protein.
For purification purposes, cells (pOTBL/HB101) were grown in a New Brunswick Bioflow (Model C300) continuous fermentor at 34 "C with a dilution factor of 0.55/h. The growth media, pH 6.8, contained M9 salts, 0.15% dextrose, 2% yeast extract, and 8% peptone. The working volume was 0.35 liter and the 0, was kept at 48%. Purification-Enzymatic activity units are given as the initial Amicromole/min in the hydrolysis of nitrocefin (0.14 mM). Protein concentrations are referred to as milligram (Bio-Rad assay) or A281 (plactamase = 29400 M" cm") (11). A typical procedure is as follows.
Harvested cell paste (116 g) was washed at 4 "C with 200 ml of T E buffer (0.1 M Tris/HCl, pH 8.0, 10 mM EDTA). Cells were centrifuged for 5 min a t 4000 X g and the wash repeated as above. The pooled supernatants, the T E washes, were concentrated using an Amicon YM-5 filter and dialyzed against imidazole buffer (25 mM, pH 7.4). The TE wash (663 mg, 160 units) contained approximately 30% of the total enzymatic activity which could easily be purified to homogeneity in high yield. The remaining activity was isolated from the pellet in low yield.
The pellet was resuspended in 100 ml of TE buffer (4 "C). Lysozyme (50 mg) was added to the solution which was then kept a t 0 "C for 30 min. Lysis was completed by freezing the suspension with dry ice followed by quick-thawing at room temperature in a water bath. The suspension was sonicated to reduce its viscosity and clarified by a 40,000 X g 1-h spin at 4 "C. The supernatant constituted the lysed (L) fraction (1800 mg, 375 units). The L-fraction required chromatography on a DEAE-cellulose column which was omitted for the TE pool. L-fraction (160 ml) was loaded on a pre-equilibrated (50 mM Tris/HCl, pH 7.2, 10 mM NaC1, 3 mM 2-mercaptoethanol, 0.01% NaN3) DEAE-cellulose (Whatman DE52) column (2.0 X 20 cm). Protein was eluted with a 0-0.2 M NaCl gradient. Fractions with enzymatic activity were pooled and dialyzed against 25 mM imidazole, pH 7.4 (120 mg, 84 units).
Protein solutions were chromatographed through a PBE 94 chromatofocusing column (Pharmacia). Protein was loaded in a 200-ml T E pool yielded 20 mg (72 units) and the L-fraction yielded 10 mg (56 units). To remove the Polybuffer and the few remaining contaminating proteins, the 8-lactamase was chromatographed through a Sephadex G-75 column (1.6 X 110 cm) in buffer (50 mM Tris/HCl, pH 7.2, 0.1 M NaCl, 3 mM 2-mercaptoethanol, 0.02% NaN3). Final yield for the TE wash was 12 mg (6 A' ,]; 90 units) and for the Lfraction 5 mg (1.75 Azsl; 36 units). The protein a t this stage of purification appeared homogenous by SDS electrophoresis and was used for the physical and kinetic studies. To obtain accurate absolute activities, a small scale Pharmacia fast protein liquid chromatography ion exchange step was utilized to remove some nonprotein UVabsorbing material and denatured enzyme. Small quantities (0.1-0.2 mg) of enzyme were chromatographed on a Pharmacia Mono Q Anion Exchanger column. The protein was eluted with a gradient (0-0.5 M ) of NaCl in 0.02 M Tris/HCl, pH 7.8, a t a flow rate of 1 ml/min.
Purified thiol 6-lactamase could be stored a t 4 "C in the presence of 2-mercaptoethanol (3 mM) over a period of weeks with little loss of activity. For more extended periods, solutions (1 mg/ml) were quickly frozen in liquid N, and stored a t -70 "C. No loss in activity was observed after thawing frozen enzyme solutions that were stored for several months.
Isolation of a True Reuertant-Cells (pIS2/HB101) were plated on ampicillin plates (100 pg/ml) a t a density of 2.106/Petri dish (100 X 15 mm). Resistant colonies were isolated at a frequency of 2 out of 10' and restreaked on ampicillin plates. Cell-free extracts prepared from these clones possessed p-chloromercuribenzoate-resistant 8lactamase activities equivalent to that given by authentic pBR322/ HB101. If one assumes a plasmid copy number of 20, the reversion rate is l.10-9. To preclude the possibility of contamination, the original plasmid pIS2 was molecularly tagged by removal of its EcoRI site (3). Plasmids from the isolated true revertants could not be cut by EcoRI.
Amino Acid Sequence Analysis-The NH2-terminal sequence of thiol /3-lactamase was determined by stepwise Edman degradation of the protein using a gas phase sequencer (Applied Biosystems, Model 470A).
Enzyme Assays-p-Lactamase activities were determined spectro- Dimethyl sulfoxide (0.5%) was presented in the assay mixture in order to solubilize nitrocefin (29). Addition of bovine serum albumin (0.25 mg/ml) and 2-mercaptoethanol (1 mM) eliminated a slow decline in the activity of thiol /3-lactamase which otherwise would occur in the nitrocefin hydrolysis assay.
The hydrolysis products of substrates were analyzed by reversed phase high pressure liquid chromatography using a DuPont Zorbax C, column (0.45 X 25 cm) and a gradient of 0-60% CH&N in 0.1% aqueous trifluoroacetic acid (30-min gradient, 1 ml/min). Hydrolysis of benzylpenicillin or nitrocefin with either wild type or thiol 8lactamase gave single products. The elution times and UV spectra of the products of these reactions were identical regardless of which enzyme was used.
Titration of Free Thiols-A solution of 5.8 nmol of thiol p-lactamase in 0.5 ml of 50 mM Tris/HCl, pH 7.0, 0.1 M NaC1, 5 mM 2-mercaptoethanol was assayed against nitrocefin. Residual reducing agents were then removed by a G-25M (Pharmacia) column (2 X 10 cm) using 50 mM Tris/HCl, pH 7.0, 0.1 M NaCl as the eluant. Fractions (0.3 ml) were collected and enzymatic aCtivity/AqSI determined. The fraction containing the most protein (1.8 nmol) had 88% of the initial specific activity. Free thiol (1.5 nmol) was determined from the which occurred on the addition of 100 nmol of 4-pyridyldi-Thiol 0-Lactamase: Purification and Properties 5329 sulfide (30). When allowance was made for the 12% loss in activity, the free thiol to active enzyme ratio was 0.99.

RESULTS
Expression of Thiol /3-Lactamase-Plasmid pIS2 which encodes thiol 0-lactamase differs from pBR322 in the one base substitution N4I0 A -T which alters codon AGC for Ser 70 of B-lactamase to that for Cys (TGC). Whereas E. coli cells hosting pBR322 are resistant to concentrations of ampicillin greater that 1 g/liter, the resistance given by pIS2 is limited to 10 mg/liter. Antibody precipitation experiments using rabbit anti-p-lactamase serum reveal that levels of thiol @-lactamase expression given by pIS2 are similar to that of wild type enzyme by pBR322. T o increase expression, we utilized the plasmid pOTBL in which a partial Tag1 restriction fragment containing the trp promotor oriented toward the thiol plactamase gene has been inserted into the Clal site of pIS2 (21).2 The trp promoter was induced by incubating cells in minimal media. After 4 h, the total cellular protein had increased 6-fold, while the cell density had only doubled. At this point, the protein band corresponding to thiol P-lactamase represented greater than 25% of the total cell protein, an approximate 100-fold enhancement (see Fig. 1). A much lower increase of 15-fold was observed for thiol p-lactamase enzymatic activity. Of this activity, 30% was present in the media. Presumably a large percentage of the enzyme occurred in an aggregated, denatured or otherwise inactive form. Although treatments might be found which will lead to higher recoveries of more active enzyme from these overexpressing cells, for initial characterization we purified enzyme from extracts where approximately 1% of the protein was active thiol 6lactamase.
Purification-The procedure established for purifying wild type p-lactamase (22) was found to be less satisfactory for thiol P-lactamase, with reduction in yield occurring during the ion exchange chromatography step. Consequently, an alternate and more rapid purification procedure was developed for purifying thiol 6-lactamase. Approximately 30% of the enzyme could be extracted from cells by washing with T E buffer. This preparation was then directly applied to a chromatofocusing column (elution at pH 5.8) followed by size exclusion gel permeation chromatography. At this point, the enzyme was pure by SDS-acrylamide electrophoresis (see Fig.   1).
Sequence Characterization-The sequence of the first 10 residues of thiol P-lactamase was determined by the automated Edman method and was found to be identical to that of mature wild type enzyme with histidine as the NHn-terminal amino acid (31). The mature wild type 8-lactamase contains 2 cysteine residues a t positions 75 and 121 linked in a disulfide bond (32). Pollitt and Zalkin (32) have shown that the reduction of this disulfide under denaturing conditions can be monitored by the difference in electrophoretic mobilities of the oxidized and reduced forms. The identical phenomenon with similar rates of reduction was observed for thiol plactamase (see Fig. 2). Titration of native thiol p-lactamase with 4-pyridyldisulfide reveals the presence of one free thiol group/enzyme molecule. The 4-pyridyldisulfide-derivatized enzyme was inactive but could be reactivated with reducing agents, e.g. 2-mercaptoethanol. In a similar manner, thiol plactamase was reversibly deactivated with p-chloromercuribenzoate.
CD Spectral Analysis-The CD spectra of thiol 6-lactamase a t 24 "C were superimposable onto that for wild type enzyme (see Fig. 3). The fractions of cy helices, p sheets, and nonrepeating structure in the proteins were estimated from least squares fits of the observed CD spectra from 204 to 243 nm to linear combinations of the reference spectra for secondary structure (33). The values found were for cy, 0.40; p, 0.20; and nonrepeating, 0.40.
Trypsin Deactiuation-While both wild type P-lactamase and thiol /3-lactamase are resistant to trypsin a t room temperature, a t higher temperatures they are cleaved to smaller peptides (31,34). The initial cleavages can be followed by the loss in p-lactamase activity. At 40 "C under the conditions described (see Fig. 4), the first order rate constants observed for the trypsin deactivations of wild type /3-lactamase and thiol p-lactamase were 0.06 and 0.02 min", respectively. The product of the wild type gene isolated as a p-chloromercuri-  The solution was kept a t 40 "C. The progress of the deactivation was monitored by removing 2-pl aliquots and assaying for nitrocefin hydrolysis activity. The amount of wild type enzyme present was determined by measuring the rate of nitrocefin hydrolysis (0.14 mM) in 1 ml of 50 mM sodium phosphate, pH 7.0, in the presence of 0.05 mM p-chloromercuribenzoate. Tot,al /3-lactamase activity was determined aft,er the addition of 2-mercaptoethanol (1 mM) and bovine serum albumin (0.15 mg/ml). A small (5%) correction in the observed rate was made to account for the spontaneous reaction of nitrocefin with the reducing media. Thiol p-lactamase activity was obtained by taking the difference in the observed total and wild type activities. 0, wild type jj-lactamase; +, thiol 8-lactamase.
benzoate-resistant revertant of the thiol p-lactamase gene possessed the same trypsin sensitivity as authentic wild type enzyme.
Enzymatic Activity-Thiol p-lactamase catalyzes the hydrolysis of representative penicillins and cephalosporins (see Tables I and 11). The observed KXn values for benzylpenicillin and ampicillin, 50 and 120 PM, respectively, are similar to those observed for wild type p-lactamase. Thiol p-lactamase kc,, values for these substrates are 1-2% that of the wild type enzyme. In the hydrolysis of nitrocefin (a cephalosporin possessing a 2,4-dinitrophenyl ring in conjugation with the 0-  Table 111. Between pH 4.5 and 9.0, the kcat for thiol P-lactamase is fairly constant as it is for wild type enzyme. The K, values for both enzymes increase at high pH. Boric Acid Inhibition-Previous workers (35) have shown that boric acid competitively inhibits serine p-lactamases with a K, of 1 mM. A small degree of noncompetitive inhibition could not be excluded. We observe that competitive inhibition Reactions were run in 50 mM buffer, 0.3 M NaCl at 24 "C. The rate constants were obtained by monitoring the reactions spectrophotometrically as described unGer "Experimental Procedures." Values for the wild type enzyme are in parentheses. Enzyme concentrations were -300 nM (thiol p-lactamase) and -3 nM (wild type 8-lactamase). In contrast to the wild type enzyme, thiol p-lactamase was not competitively inhibited by boric acid. A small noncompetitive inhibition was observed a t high boric acid concentrations. Boric acid at 10 mM inhibited thiol p-lactamase hydrolysis of benzylpenicillin by 12-16% for substrate concentrations of 0.05-0.5 mM and inhibited thiol p-lactamase hydrolysis of nitrocefin (0.14 mM) by 15%.

DISCUSSION
The primary object of the present work was to determine whether the chemical behavior of an enzyme, in this case pBR322 8-lactamase, could be manipulated by specific changes in its primary sequence. Determining the consequences of specific changes sheds light on structure-function relationships which in turn will direct future changes that will produce new proteins with desired properties. We have already described the manner by which the codon for the active site Ser 70 of p-lactamase was changed to that of Cys (7). In the present study, this new protein, thiol p-lactamase, was purified and characterized.
Structurally, thiol p-lactamase is similar to wild type plactamase. The disulfide bond (Cys 75"Cys 121) in both enzymes is reduced under denaturing conditions at approximately equivalent rates. At 24 "C, the CD spectra of both enzymes are nearly identical. These spectra reflect a high degree of secondary structure with 40% cy helix and 20% p sheet being predicted by the method of Chen et al. (33).
Similar predictions of helical content and CD spectra have been reported for Bacillus cereus and Staphylococcus aureus p-lactamases (36). A crystal structure remains to be determined for a p-lactamase. In view of the strong homology between known p-lactamases and D-alanykarboxypeptidases, it should be noted that the structure of one serine D-alanylcarboxypeptidase is known and shows a high percentage of a helix and /j sheet (37).
Although thiol p-lactamase and wild type enzyme appear to have similar structures, they differ in their sensitivity toward trypsin. Denatured wild type enzyme is rapidly and extensively digested by trypsin (31). However, the native enzyme is fairly resistant to trypsin and other proteases and is cleaved only at high (40 "C) temperatures. In contrast, several thermolabile 8-lactamases as well as the precursor form are extremely sensitive to trypsin (0 "C) (34). At 40 "C, thiol p-lactamase is %fold more resistant to trypsin than wild type enzyme. This translates to a difference of 0.6 kcal/mol (RT In 3) between the energies of activation for cleavage of the two enzymes. If we assume that trypsin cleaves only a non-native form of both enzymes and that the rate of this cleavage is proportional to the population of the non-native state, then the Cys to Ser substitution has resulted in a 0.6kcal/mol stabilization of the native conformation of @-lactamase. The magnitude and direction of this change are consistent with replacing a hydroxyl with a sulfhydryl group in the apolar interior of a protein. Sulfur is larger (van der Waals radii 1.8 A for 0), softer, and more polarizable than oxygen (38). Fersht and Dingwall (39) have discussed the difference in hydrophobicity of the -SH and -OH groups and the role it might play in the selectivity of tRNA synthetases. They calculated a 1.5 kcal/mol of more favorable dispersion energy for the interaction of -Swith a -CH2-group than for -0-.
Of particular interest was the difference in enzymatic activity between wild type p-lactamase and thiol p-lactamase. The enzymatic catalyzed hydrolysis of p-lactams is a multistep process in which an acyl-enzyme intermediate forms and then is hydrolyzed. The formation of the acyl-enzyme intermediate is itself a complicated series of chemical steps each of which will be influenced by the -OH to -SH substitution. Model studies demonstrate that thiols are more effective nucleophiles than alcohols for displacement reactions at carboxyl groups with good leaving groups, e.g. p-nitrophenoxide or imidazolium ions (41,42). In these studies, the nucleophilic species were the ionized thiolate and alkoxide anions. A large dependence on the leaving group for thiolate nucleophiles might result from competition in the partitioning of the tetrahedral intermediate to product as opposed to reactants (here the thiolate anion being the leaving group). The breakdown of the tetrahedral intermediate to form reactants would occur less rapidly with alkoxides which are poor leaving groups.
In view of these model studies, it is not surprising that thiol-subtilisin hydrolyses activated esters such as p-nitrophenylacetate nearly as rapidly as does subtilisin (17,18). Furthermore, the rate-determining deacylation for thiol-subtilisin was found to be 10% that for subtilisin which is consistent with the greater stability of thiol esters toward general acid-and base-catalyzed hydrolysis (43). However, thiol-subtilisin lacks activity toward amide substrates although sulfhydryl proteases such as papain and ficin do exist. Previously, this lack of reactivity had been attributed to incorrect orbital orientation (17). Alternatively, in the formation at the acyl-enzyme intermediate, the stabilization of the tetrahedral intermediate and the rapid protonation of the amine leaving group might be more critical for cysteine pro-Thiol 0-Lactamase: Purification and Properties teases than for serine proteases.
The situation with (3-lactam antibiotics is somewhat intermediate between that of activated esters and amides. The bicyclic ring system found in penicillins and cephalosporins forces the p-lactam amide group to be nonplanar and suppresses the amide reasonance -CO-N'H- (44). @-Lactam antibiotics are therefore activated amides which are more easily hydrolyzed. An excellent organic model for a p-lactamase is, in fact, ionized cysteine which reacts with benzylpenicillin with a rate three times that of hydroxide anion (45).
As this study shows, thiol /3-lactamase has appreciable filactamase activity. For benzylpenicillin kcat/& is 3.5 X lo5 s-' M" or 1% of the wild type 4 X lo7 s-' M-'. This bimolecular constant is lo5 greater than that for the reaction of cysteine with benzylpenicillin at pH 7.0 (45). This enhancement in rate over that observed for model systems consisting simply of the chemical catalytic groups results from an unknown combination of factors such as substrate distortion, microenvironment, and orbital orientation which have been proposed to account for enzymatic catalysis (46,47). The pH dependence for thiol @-lactamase further supports our belief that we are observing a true enzymatic reaction. Whereas in the cysteine hydrolysis of penicillins the ionized thiolate form is the active species, the activity of thiol p-lactamase is actually greater at low pH values. The small differences between the pH profiles for thiol p-lactamase and wild type enzyme at the low and high pH extremes might, upon further investigation, give useful mechanistic information.
Most noteworthy was thiol p-lactamase's unusual reactivity toward nitrocefin (a cephalosporin with an electron-withdrawing 2,4-dinitrophenyl ring in conjugation with the plactam N). The kc,, for this substrate was at least as high as for the wild type enzyme although the k,,,/K,,, remained at 2% of the wild type value. The large kc,, increase observed with an electron-withdrawing substituent is reminiscent of the results with thiol-subtilisin. The drastic change in activity (kcat) upon alteration of the 0-lactam electronic character strongly argues for the rate-determining step being the formation of the acyl-enzyme intermediate. Unfortunately, with good substrates, the rate-determining step is not known for wild type 8-lactamase. Much more mechanistic detail is available from studies with poor substrates (11). Similar studies with thiol p-lactamase should enable the comparison of the individual rate constants for the formation of the intermediate and its breakdown between the two enzymes.
The disappearance of boric acid competitive inhibition upon Ser 70 to Cys 70 substitution demonstrates the ability of specific mutations to solve protein structure-function problems. Previous workers (35) attributed boric inhibition of @lactamase to the same covalent complex involving the active site serine as observed by crystallography for subtilisin. The absence of this inhibition with thiol P-lactamase substantiates this mechanism.
In summary, the enzyme that results from the Ser 70-Cys 70 substitution of pBR322 P-lactamase, thiol p-lactamase, has different physical and enzymatic properties. As such, this enzyme should prove useful in studies of the physiology (secretion and processing), the folding, and the enzymology of 6-lactamases.